Have you ever let an AI agent run commands on your production server because you were stuck, telling yourself it would be fine?
In the latest episode of the No Compromises podcast, we discuss why giving AI agents access to your production environment is as reckless as the bad old days of FTPing code straight to your server.
We make the case that AI agents are non-deterministic and carry no responsibility, which makes them far more dangerous than a developer making a manual mistake on a live server.
We also cover why you should never blindly run complex chained commands that an agent suggests, how to verify what your agent is actually doing, and why a little friction in your workflow is worth the safety it buys you.
- (00:00) - FTPing to production is still wrong in 2026
- (01:50) - Running Tinker on production is no better
- (02:28) - AI agents on production servers is even worse
- (07:11) - How to verify what your agent is actually doing (move from 07:00)
- (11:03) - Silly bit
Episode 157
